1:
2: \documentclass[showpacs,prl,twocolumn,aps]{revtex4}
3: \usepackage{bm}
4: \usepackage{amsmath}
5: \usepackage{amssymb}
6: \usepackage{latexsym}
7: \usepackage{amsfonts}
8: \usepackage{epsfig}
9: \usepackage{psfrag}
10:
11: \newcommand{\kf}{{\boldsymbol k}}
12: \newcommand{\ket}[1]{\left|#1\right>}
13: \newcommand{\bra}[1]{\left<#1\right|}
14: \newcommand{\braket}[2]
15: {\left<#1|#2\right>}
16: \newcommand{\nn}{\nonumber\\}
17: \newcommand{\ul}{\underline}
18: \newcommand{\f}[1]{\mbox{\boldmath$#1$}}
19: \newcommand{\fk}[1]{\mbox{\boldmath$\scriptstyle#1$}}
20: \newcommand{\vau}{\mbox{\boldmath$v$}}
21: \newcommand{\na}{\mbox{\boldmath$\nabla$}}
22: \newcommand{\bea}{\begin{eqnarray}}
23: \newcommand{\ea}{\end{eqnarray}}
24: \newcommand{\eea}{\end{eqnarray}}
25: \newcommand{\ord}{{\cal O}}
26: \newcommand{\sumint}[1]
27: {\begin{array}{c}
28: \\
29: {{\textstyle\sum}\hspace{-1.1em}{\displaystyle\int}}\\
30: {\scriptstyle{#1}}
31: \end{array}}
32:
33: \begin{document}
34:
35: \title{Stability %and existence
36: of quasi-two-dimensional
37: Bose-Einstein condensates \\ with dominant dipole-dipole interactions}
38: % \\with dipole-dipole interactions}
39:
40: \author{Uwe R.~Fischer}
41:
42: \affiliation{Eberhard-Karls-Universit\"at T\"ubingen,
43: Institut f\"ur Theoretische Physik\\
44: Auf der Morgenstelle 14, D-72076 T\"ubingen, Germany}
45:
46: \begin{abstract}
47: We consider quasi-two-dimensional atomic/molecular Bose-Einstein condensates with both contact and dipole-dipole interactions.
48: It is shown that, as a consequence of the dimensional reduction,
49: and within mean-field theory, the condensates do not develop unstable
50: excitation spectra,
51: % up to large densities (in comparison with three-dimensional condensates),
52: even when the dipole-dipole interaction completely dominates
53: the contact interaction.
54: \end{abstract}
55:
56: \pacs{03.75.Kk, 03.75.Hh} %, 05.30.Jp}
57:
58: \maketitle
59:
60: Dilute Bose-Einstein condensates with long-range interactions
61: offer promising opportunities to explore
62: the potentially strong
63: correlations induced by the interaction, which go beyond the
64: comparatively weak correlations %ones
65: induced by the local
66: contact interaction pseudopotential conventionally
67: sufficient to describe most atomic condensates.
68: The recent realization of a concrete physical system
69: in which atomic magnetic dipoles play a significant role \cite{Griesmaier},
70: %, which is however still weak as compared to the contact interaction
71: % coupling,
72: is a first experimental step towards the exploration of dipolar condensate
73: physics \cite{Goral,YiYouII,SantosZoller,ODell,Baranov}.
74: To investigate {\em dipole-dominated} physics, greater potential than by
75: magnetic dipoles is offered by heteronuclear molecules with an electric dipole
76: moment \cite{YiYou,Kotochigova}, due
77: to the fact that the electric dipole
78: interaction strength is larger than the magnetic one by a
79: factor of about $10^4$ for typical atomic/molecular
80: dipole moments of 1 Bohr
81: magneton or 1 Debye, respectively.
82:
83: An interesting property of three-dimensional (3D) %1D)
84: condensates is the existence of a ``roton'' minimum, which stems from the
85: fact that the Fourier transform of the dipole-dipole
86: interaction, occurring in the Bogoliubov mean-field spectrum,
87: assumes negative values in certain directions in momentum space
88: \cite{RotonDipolar,RotonsDipole,EPJD}.
89: This property however also causes a problem
90: of 3D dipolar condensates, as they become %{\em necessarily}
91: dynamically unstable for dipole-dipole interactions dominating the contact
92: interaction, at densities corresponding to the Thomas-Fermi limit of
93: %sufficiently
94: large condensates.
95: The roton minimum quickly deepens with increasing density and/or
96: dipole coupling to hit the zero of energy, and a further increase
97: is impossible because the system develops imaginary excitation energies.
98:
99: In the present study an analytical proof is given that,
100: in contrast to 3D condensates,
101: quasi-two-dimensional (quasi-2D)
102: dipolar condensates are stable, %up to comparatively large densities,
103: even when the dipole coupling completely dominates the contact coupling.
104: The experimental realization of purely dipolar condensates is thus
105: possible only in the quasi-2D regime. The physical reason
106: for the stabilization is that dimensional reduction entails
107: that the mutual dipole-dipole interaction of the atoms/molecules,
108: with dipoles oriented perpendicular to the
109: plane in the strongly confined direction, can effectively sample
110: an exponentially smaller region in configuration space where the interaction
111: assumes negative values.
112: %As a consequence, it is possible to approach,
113: %in a controlled manner, that is
114: %by reducing the trapping in the strongly confined
115: It is, furthermore, shown that a
116: strongly correlated regime of purely dipolar condensates,
117: where large quantum depletion of the condensate occurs
118: and the mean-field description in terms of a single condensate
119: wave function breaks down, takes place in a
120: crossover regime from quasi-2D to 3D.
121: For an electric-dipole-dominated dilute gas,
122: %this %the strongly correlated
123: %regime will presumably be {\em necessarily} reached
124: %in experiment:
125: the densities corresponding to this crossover regime to
126: 3D turn out to be quite small,
127: even within the strongly enlarged stability window of
128: strongly anisotropic pancake-shaped condensates.
129: % for atomic dipole moments of the order of Debye.
130:
131: The mean-field description of the system we start with is based
132: on a nonlocal 3D Gross-Pitaevski\v\i\/ equation for the order
133: parameter $\Psi$
134: (we put $\hbar = m =1$, where $m$ is the mass of the atoms/molecules)
135: \cite{Goral,YiYou}:
136: % comment on that assumption later on
137: \begin{eqnarray}
138: i \frac\partial{\partial t} \Psi ({\bm r}, t)
139: & = & \left\{ -\frac{\nabla^2} 2 +\frac12 \omega_z^2 z^2 +
140: g_{\rm 3D} |\Psi ({\bm r},t)|^2\right. \label{GP}\\
141: %& &
142: %\hspace*{6em}
143: & &\left. + \int d^3r'
144: V_{dd}({\bm r}-{\bm r}') |\Psi ({\bm r}',t)|^2 \right\}\! \Psi ({\bm r}, t) .\nonumber
145: %\nonumber\\
146: \end{eqnarray}
147: The trapping in the plane is assumed to be
148: negligible as compared with the strong confinement,
149: of harmonic trapping frequency $\omega_z$, in the $z$ direction.
150: The term in the second line contains the dipole-dipole interaction of the
151: atoms/molecules, with all dipoles oriented by an %externally
152: applied field along the strongly confining $z$ direction:
153: \begin{equation}
154: V_{dd} ({\bm r}) = \frac{3g_d}{4\pi} \frac{1-3z^2/|{\bm r}|^2}
155: {|{\bm r}|^3} . \label{Vdd}
156: \end{equation}
157: The dipole coupling reads $g_d= \mu_0 d_m^2/3 $ for magnetic and
158: $g_d=d_e^2/3\epsilon_0 $ for electric dipoles. In the units
159: %with $\hbar = m= 1$
160: which we employ, the $g_d$ are having dimensions of length like
161: the 3D contact interaction $g_{\rm 3D}=4\pi a_s$, for
162: which the length scale is set by the $s$-wave scattering length $a_s$.
163: We note that the value of the contact interaction coupling strength
164: generally depends on the value of $g_d$
165: (in an effective-dimension dependent manner),
166: because the long-range dipole-dipole interaction affects the
167: short-range scattering processes \cite{YiYouII}.
168: For the statements made in the present paper,
169: the actual ratio $g_{\rm 3D}/g_d$ and the absolute
170: value of the (unscreened) $g_d$ will be relevant.
171:
172: For a quasi-2D Bose-Einstein condensate,
173: the motion of the atoms/molecules is
174: by definition restricted to zero point oscillations
175: in a harmonic oscillator potential.
176: %The condensate density profile
177: %in the strongly confining $z$ direction is then simply
178: %the Gaussian of the harmonic oscillator ground state.
179: We then take as a general ansatz for the density %profile
180: %($\rho({\bm r})=|\Psi({\bm r})|^2$)
181: \begin{equation}
182: \rho({\bm r}) = |\Psi({\bm r})|^2 =\frac1{\sqrt{\pi d_z^2}}
183: \exp\left[-\frac{z^2}{d_z^2}\right] n(x,y) ,
184: \label{quasi2D}
185: \end{equation}
186: where the density in the plane, $n(x,y)$, is normalized to the
187: total number of particles $N$.
188: We treat $d_z$ as a parameter minimizing the
189: Gross-Pitaevski\v\i\/ ground state energy \cite{YiYou,EPJD}.
190: Assuming homogeneous density in the plane, the equation determining
191: $d_z$ reads $\omega_z^2= d_z^{-4}
192: + {(g_{\rm 3D} + 2g_d)} nN d_z^{-3}/{\sqrt{2\pi}}$.
193: If the right-hand side is dominated by the first (kinetic energy) term,
194: $d_z$ equals the harmonic oscillator length
195: $1/\sqrt{\omega_z}$, the system is quasi-2D,
196: % such that the average energy per particle
197: %given by the chemical potential is much less than $\omega_z$;
198: and the above Gaussian gives
199: the density profile exactly.
200: %and increases with an increase in effective dimension.
201: Defining a parameter $\alpha \equiv \omega_z d_z^2$, we have
202: $\alpha =1$ if the system is quasi-2D and $\alpha \gg 1$
203: deep into the 3D regime, where $\alpha$ becomes interaction dependent.
204:
205: We now calculate the total dipole-dipole energy given that the
206: density profile in $z$-direction is prescribed by the above Gaussian.
207: In accordance with Eq.\,(\ref{GP}),
208: the Gross-Pitaevski\v\i\/ energy functional
209: of the dipole-dipole interaction generally reads
210: \begin{eqnarray}
211: H_{dd} &= & \frac12 \int d^3r \int d^3r' \rho({\bm r}) V_{dd}({\bm r}-{\bm r}')
212: \rho({\bm r}') \nonumber\\
213: & = & \frac12 \frac1{(2\pi)^3} \int d^3k \tilde \rho ({\bm k})
214: \tilde V_{dd} ({\bm k}) \tilde \rho (-{\bm k}),
215: \end{eqnarray}
216: where the second line employs a convolution to Fourier space.
217: The Fourier transform of the dipole-dipole interaction (\ref{Vdd})
218: takes the form
219: $\tilde V_{dd} ({\bm k})=g_d [3k_z^2/(k_x^2+k_y^2+k_z^2)-1]$;
220: using (\ref{quasi2D}), we obtain the Fourier transform of the density
221: $\tilde \rho({\bm k})= \exp\left[-\frac14 k_z^2 d_z^2\right]\tilde n (k_x,k_y)$.
222: Integrating over the $k_z$ direction,
223: the effective dipole-dipole energy is then given by
224: a two-dimensional integral in $(k_x,k_y)$ space
225: \begin{eqnarray}
226: H_{dd}
227: %&= & \frac12 \frac1{(2\pi)^3} \int d^2k \tilde n (k_x,k_y)
228: %\tilde n (-k_x,-k_y) \nonumber\\
229: %& &\times g_d \int_{-\infty}^{+\infty} dk_z
230: %\exp\left[-\frac12 k_z^2 d_z^2\right]\left(\frac{3k_z^2}{k_x^2+k_y^2+k_z^2}-1\right)
231: %\nonumber\\
232: & = & \frac{g_d}2 \frac1{(2\pi)^2} \int d^2k \tilde n (k_x,k_y)
233: \tilde n (-k_x,-k_y) \nonumber\\
234: & & \times \left\{ \frac{2}{\sqrt{2\pi} d_z}
235: -\frac{3}2 \exp\left[\frac{k^2 d_z^2}2\right] k\, {\rm Erfc} \left(\frac{kd_z}{\sqrt 2}\right)\!
236: \right\}
237: \label{Hdd}
238: \end{eqnarray}
239: where the complementary error function
240: Erfc $(z) = 1-{\rm erf}(z) = 1-(2/\sqrt{\pi}) \int_0^z \exp(-t^2) dt$
241: and $k=(k_x^2+k_y^2)^{1/2}$; $\tilde n(k_x,k_y)$ is the Fourier transform of
242: the 2D density.
243: Employing the same procedure of integrating out the Gaussian in the
244: $z$ direction, for the contact interaction part in the
245: Gross-Pitaevski\v\i\/ energy functional, yields the well-known result for
246: the 2D effective coupling $g_{\rm 2D} = 2\sqrt{2\pi}
247: a_s/d_z$ (valid in the limit that the 3D $s$-wave scattering length
248: $a_s\ll d_z$
249: \cite{Holzmann}).
250: \begin{center}
251: \begin{figure}[t]
252: \psfrag{Vk}{\large $\tilde V^{\rm 2D}_{\rm tot}$} %[\rho(0)d_z^3/A]$}
253: \psfrag{kz}{\large $kd_z$}
254: \centerline{\epsfig{file=Fig1.eps,width=0.385\textwidth}}
255: \caption{\label{Fourier}
256: Fourier transform of the total interaction potential
257: in units of $A/\rho(0)\sqrt{\pi}d_z^3$, as a function of
258: $\zeta =k d_z$. The values of $R=\frac12 \sqrt{\pi/2}$ (solid line),
259: $R=\frac23 \sqrt{\pi/2} $ (dotted line)
260: $R=\sqrt{\pi/2}$ (dipole interaction dominating; dashed line). For
261: $R=\frac23 \sqrt{\pi/2}$ (i.e., when $g_{d}= g_{\rm 3D}$),
262: the Fourier transform approaches zero at infinite
263: momentum.}
264: \end{figure}\vspace*{-2.5em}
265: \end{center}
266:
267: From the relation (\ref{Hdd}) for the dipole-dipole contribution
268: in the interaction energy, it follows that the Fourier transform of the
269: total, contact plus dipole, interaction potential assumes the form
270: \begin{equation}
271: \tilde V^{\rm 2D}_{\rm tot} ({\zeta})
272: = \frac A{\rho(0)\sqrt{\pi} d_z^3}\left\{
273: 1-\frac{3R}{2}\, \zeta\, w\left[\frac{\zeta}{\sqrt 2}\right]
274: %\exp\left[\frac12 \zeta^2\right] {\rm Erfc} \left[\frac{\zeta}{\sqrt 2}\right]
275: \right\},
276: \label{V2D}
277: \end{equation}
278: where we made use of the $w$-function, related to Erfc$(z)$ by
279: $w(z)= \exp[z^2]{\rm Erfc}(z)$ \cite{Abramowitz,note}.
280: The dimensionless wavenumber ${\zeta} = k d_z $, and the two dimensionless parameters
281: occurring in $\tilde V^{\rm 2D}_{\rm tot} ({\zeta})$, using the central
282: 3D density $\rho(0)$, are defined to be
283: \begin{equation}
284: A = \frac{\rho(0)\sqrt{\pi}d_z^2 g_d}R \,,\qquad
285: %\sqrt{2} \rho(0){d_z^2}\left(\frac{g_{\rm 3D}}2 + {g_d}\right)\,,\qquad
286: R = \frac{\sqrt{\pi/2}}{1+g_{\rm 3D}/2g_d}. \label{Adef}
287: \end{equation}
288: The value of the parameter $R$ ranges from $R=0$
289: if $g_d/g_{\rm 3D} \rightarrow 0$, to $R=\sqrt{\pi/2}$ for
290: $g_d/g_{\rm 3D}\rightarrow \infty$.
291: The second term in the curly brackets in Eq.\,(\ref{V2D})
292: rapidly decreases as a function of $\zeta$ and approaches a constant,
293: which is due to the fact that $w(\zeta/\sqrt2) \rightarrow \sqrt{2/\pi} \,
294: \zeta^{-1}$ for $\zeta\rightarrow \infty$,
295: cf. the plot of $\tilde V^{\rm 2D}_{\rm tot} (\zeta)$ in Fig.\,\ref{Fourier}.
296: For small $\zeta$, the
297: quasi-2D Fourier transform behaves like
298: $\tilde V^{\rm 2D}_{\rm tot} ({\zeta})
299: = \frac A{\rho(0)\sqrt{\pi} d_z^3}
300: \left[1-\frac{3R}{2}{\zeta} +\frac{3R}{\sqrt{2\pi}}\zeta^2 +
301: {\cal O} (\zeta^3)\right]$.
302: Observe that $\tilde V^{\rm 2D}_{\rm tot} (\zeta)$
303: %is {\em regular}
304: possesses a well-defined value at the origin $\zeta=k=0$,
305: as opposed to the 3D Fourier transform of the
306: dipole-dipole interaction potential.
307:
308:
309:
310:
311:
312:
313: From the Fourier transform of the
314: interaction %Eq.\,
315: (\ref{V2D}), we conclude that the squared Bogoliubov spectrum
316: \cite{Bogoliubov,Foldy},
317: for excitations confined to the plane,
318: $\omega^2 =
319: \rho(0)\sqrt{\pi} d_z \tilde V^{\rm 2D}_{\rm tot} (k) k^2 + k^4/4$
320: is, in units of
321: $1/d_z^4$, given by
322: \begin{equation}
323: \epsilon^2(\zeta)=
324: A\zeta^2 \left( 1 - \frac{3R}2
325: \zeta\, w\left[\frac{\zeta}{\sqrt 2}\right]
326: %\exp\left[\frac12 \zeta^2\right]{\rm Erfc}
327: %\left[\frac{\zeta}{\sqrt 2}\right]
328: \right)
329: + \frac{\zeta^4}4. \label{Bogoliubov}
330: \end{equation}
331: The main observation of the present analysis is that,
332: as opposed to the 3D case (or, in an exacerbated manner,
333: the quasi-1D case \cite{EPJD}), the %quasi-2D
334: Bogoliubov spectrum (\ref{Bogoliubov}) does not necessarily become unstable
335: if the dipole interaction coupling exceeds the contact interaction
336: coupling.
337: %for $g_{\rm 3D}/g_d\rightarrow 0$. %, i.e. for $R\rightarrow\sqrt{\pi/2}$.
338: Assuming $g_{\rm 3D}/g_d \rightarrow 0$, i.e. $R\rightarrow\sqrt{\pi/2}$,
339: the above squared spectrum can assume negative values, and hence the
340: excitation energies become imaginary, when %the dimensionless coefficient
341: $A$ exceeds the critical value
342: $A_c = 3.446 $, cf. Fig.\,\ref{Spectrum}, where we show a sequence of
343: four spectra for different $A$ at constant $R=\sqrt{\pi/2}$.
344:
345: The critical value of $A_c=3.446$ corresponds to a critical dipole coupling
346: given by $({g_d)}_c = {2.436}/{\rho(0) d_z^2}$.
347: Using numbers appropriate for atomic magnetic moments $\mu_m = N_m \mu_B$,
348: we find that $g_d = 5.4\times 10^{-6}\, \mu$m\,$M N_m^2$
349: ($M$ is the mass of the atoms or molecules in units of the
350: atomic mass unit $= 1.66\times 10^{-27}$\,kg). In the case
351: of electric dipoles, with moment $d_e= N_e$\,Debye,
352: we find $g_d = 6.2\times 10^{-2} \,\mu$m\,$MN_e^2$.
353: The critical coupling $({g_d)}_c$
354: for dominant dipole-dipole interactions
355: then translates into a critical central
356: 3D density
357: %\begin{eqnarray}
358: $
359: \rho_c(0) = 4.5\times 10^{16} {\rm cm}^{-3} N_m^{-2} \alpha^{-1}
360: \omega_z [2\pi \times
361: {\rm kHz}]
362: $
363: and
364: %\nonumber\\
365: $
366: \rho_c(0) = 3.8\times 10^{12} {\rm cm}^{-3} N_e^{-2} \alpha^{-1}
367: \omega_z [2\pi \times {\rm kHz}]
368: $
369: %\end{eqnarray}
370: in the case of magnetic and electric dipoles, respectively.
371: %%
372: \vspace*{-1.25em}
373: \begin{center}
374: \begin{figure}[t]
375: \psfrag{E}{\large $\epsilon^2$}
376: \psfrag{kz}{\large $kd_z$}
377: \centerline{\epsfig{file=Fig2.eps,width=0.385\textwidth}}
378: \caption{\label{Spectrum}
379: Squared Bogoliubov spectrum
380: in units of $1/d_z^4$, Eq.\,(\ref{Bogoliubov}), where
381: $R=\sqrt{\pi/2}$ ($g_{\rm 3D}/g_d \rightarrow 0$) for all curves.
382: Counting from bottom to top at small $kd_z$ the values of $A$ in
383: Eq.\,(\ref{Adef}) are:
384: $A=A_c/10$, $A=A_{\rm min}$, $A=A_c$, and $A= 1.2 A_c$.
385: For $A>A_{\rm min}=1.249$, the spectrum displays a roton minimum, and becomes
386: unstable for $A>A_c=3.446$.}
387: \end{figure}\vspace*{-1.5em}
388: \end{center}
389:
390: The quantity $\sqrt{\pi} \rho(0) g_d/R$, a measure of the
391: energy per particle (the chemical potential),
392: is equal to $A/d_z^2$. For the system to be
393: quasi-2D, we therefore need $A\ll \omega_z d_z^2 =\alpha \equiv 1$.
394: %(in quasi-2D).
395: Furthermore, the Bogoliubov spectrum in Fig.\,\ref{Spectrum}
396: does not possess any points where $d\epsilon^2/d\zeta^2$=0
397: %, i.e., is essentially of a (renormalized, cf.\,Eq.(\ref{}))
398: %contact interaction form,
399: for values $A< A_{\rm min} = 1.249 $ (when $R=\sqrt{\pi/2}$). That is,
400: a ``roton'' minimum cannot develop within the regime of quasi-2D.
401: %, and up to large values of $g_d$ and/or central density.
402: Thus we conclude that a quasi-2D purely dipolar system of bosons is
403: always stable, and that the instability of the
404: condensate takes place in the crossover region to 3D.
405: By contrast, deep inside the 3D regime, $g_{\rm 3D} \ll g_d$
406: implies collapse of the condensate for any reasonable value of the density,
407: and in the Thomas-Fermi limit
408: the condensate will be unstable for any $g_d$ slightly exceeding $g_{\rm 3D}$
409: \cite{RotonDipolar}.
410:
411: The critical value of $A_c(R)$, at which the gas becomes unstable, obtained
412: by numerically finding the large momentum zeros of the Bogoliubov spectrum
413: (\ref{Bogoliubov}), exponentially increases for smaller $R$
414: (increasing $g_{\rm 3D}/g_d$), and
415: diverges at $R=\frac23 \sqrt{\pi/2}$ (i.e., for $g_{\rm 3D}=g_d$),
416: where the Fourier transform $\tilde V^{\rm 2D}_{\rm tot}$ becomes positive
417: everywhere. %, cf.\,Fig.\,\ref{Fourier}.
418: Comparing the critical values of $A$ thus obtained
419: to those from a full 3D solution of the
420: Bogoliubov-de Gennes equations \cite{RotonDipolar},
421: the present approach reproduces the critical $A$ sufficiently
422: accurate in the dipole-dominated case,
423: %(to within about ten to twenty percent),
424: for which the instability takes
425: place in the crossover regime from quasi-2D to 3D.
426: Deep into the 3D regime, %where $A\gtrsim O(10)$,
427: the critical value $A_c$ calculated from (\ref{Bogoliubov})
428: is strongly overestimated, mainly because
429: it is exponentially sensitive on the exact form of the spectrum.
430:
431:
432:
433:
434:
435:
436: %Let us now discuss the consequences of the spectrum (\ref{Bogoliubov})
437: %for the stability of the condensate with
438: %numbers appropriate to experiment.
439:
440:
441: %We conclude that in the quasi-2D case it is possible to produce
442: %strongly correlated systems by virtue of the long-range nature of
443: %purely dipolar condensates
444: %with a typical range of the potential approaching
445: %the interparticle separation,
446: %without endangering their stability by imaginary excitation energies.
447: %In particular, in the electric case, the critical range of the
448: %dipole potential may even exceed the interparticle
449: %separation by an order of magnitude.
450:
451:
452: At nonzero temperature $T$, 2D Bose-Einstein condensates
453: do not exist in the homogeneous case \cite{Hohenberg},
454: while trapping in the plane
455: enables the existence of a condensate also at finite $T$ \cite{Fischer}.
456: On the other hand, in the presently discussed $T=0$ case,
457: it is a well-established fact
458: that even without trapping, Bose-Einstein
459: condensation occurs in two spatial dimensions,
460: see, e.g.,\,\cite{PitaLongRange}.
461: We next turn to a discussion of the zero temperature value of the number
462: density of excitations above the condensate, the so-called quantum depletion.
463: To this end, we use a
464: mode expansion for the annihilation operators $\hat\chi_\kf$
465: of the original bosons in terms of the Bogoliubov
466: quasiparticle operators $\hat a_{\bm k}, \hat a^\dagger_{\bm k}$
467: \cite{Schutzhold}:
468: \begin{equation}
469: \label{creation}
470: \hat\chi_{\boldsymbol k}
471: = \sqrt{\frac{{\boldsymbol k}^2}{2\epsilon_{\boldsymbol k}}}
472: \left[ \left(\frac12+\frac{\epsilon_{\boldsymbol k}}{{\boldsymbol k}^2}\right)
473: \hat a_{\boldsymbol k}
474: +\left(\frac12-\frac{\epsilon_{\boldsymbol k}}{{\boldsymbol k}^2}\right)
475: \hat a_{\boldsymbol k}^\dagger
476: \right].
477: \end{equation}
478: The above form of the Bogoliubov transformation
479: results, after inversion, in the usual phonon quasiparticle
480: operators at low momenta, and gives
481: $\hat \chi_{\bm k} = \hat a_{\bm k}$ at ${\bm k}\rightarrow \infty$, i.e.,
482: the quasiparticles and the bare bosons become, as required,
483: identical at large momenta.
484:
485: The quantum depletion density at zero temperature %density
486: is calculated by evaluating the expectation of $\hat\chi^\dagger\hat\chi$
487: in the quasiparticle vacuum defined by
488: $\hat a_{\bm k}|{\rm vac}\rangle=0$:
489: \bea
490: \label{dipole}
491: \langle %{\rm vac}|
492: \hat\chi^\dagger\hat\chi %|{\rm vac}
493: \rangle
494: &=& \frac1{2\pi d_z^2} \int_0^\infty d\zeta \, \frac{\zeta^3}{2\epsilon(\zeta)}
495: \left(\frac12 -\frac{\epsilon(\zeta)}{\zeta^2}\right)^2
496: .
497: \ea
498: The instability generally happens because {\em in-plane}
499: excitations become of imaginary frequency \cite{RotonDipolar};
500: above the calculated critical value of $A=A_c$,
501: the in-plane excitations have vanishing energy, and
502: the quantum depletion % according to Eq.\,(\ref{dipole})
503: diverges.
504: Using the in-plane momentum integral above to calculate the depletion, with
505: the effective dispersion relation (\ref{Bogoliubov}),
506: assumes that the corrections due to the
507: neglected out-of-plane Bogoliubov excitations,
508: characterized by transverse quantum numbers,
509: are small. This is justified because
510: the dominant contribution to quantum depletion is, for a dilute Bose gas,
511: from large momentum excitations with (approximately)
512: vanishing energy.
513: %, irrespective of the (effective) dimension.
514: Out-of-plane excitations have large energies at the relevant in-plane
515: momenta of order $1/d_z$; they thus
516: do not contribute significantly to the depletion.
517: %[the factor in
518: %front of the round brackets in the integral of Eq.\,(\ref{dipole}) becomes
519: %$\zeta^{D+2}/2\epsilon$ in D spatial dimensions].
520: In Fig.\,\ref{depl}, we show the result for the quantum
521: depletion in the case of
522: %two different values of $R$, corresponding to
523: dipole-interaction dominated condensates, for which we can expect the
524: in-plane spectrum (\ref{Bogoliubov}) to be sufficiently
525: accurate up to $A\simeq A_c$.
526: % and condensates with comparable dipole
527: %and contact-interaction coupling constants.
528: %For $R\rightarrow 0$, that is for dominant contact interaction,
529: %the normalized depletion becomes, as
530: %expected, linear in $A$,
531: %$d_z^2 \langle\hat\chi^\dagger\hat\chi\rangle= A/4\pi=
532: %\rho(0) d_z^2 g_{\rm 3D}/4\sqrt2 \pi $ (valid for $A\ll 1$).
533: At the critical value $A=A_c(\sqrt{\pi/2})\simeq 3.4$
534: %, which is for $R=\sqrt{\pi/2}$ located at $A_c=3.446$
535: (dotted vertical line in Fig.\,\ref{depl}),
536: the condensate depletion diverges, and the mean-field
537: condensate will yield to a new quantum phase.
538:
539: \begin{center}
540: \begin{figure}[t]
541: \psfrag{Depl}{\normalsize $d_z^2\langle\hat\chi^\dagger\hat\chi\rangle$}
542: \psfrag{A}{\normalsize $A$}
543: \centerline{\epsfig{file=Fig3.eps,width=0.34\textwidth}}
544: \caption{\label{depl}
545: Quantum depletion density from Eq.\,(\ref{dipole}),
546: %in units of $d_z^{-2}$,
547: as a function of $A=\rho(0) \sqrt{\pi} d_z^2 g_d /R$,
548: for $R=\sqrt{\pi/2}$ ($g_d/g_{\rm 3D} \rightarrow \infty$).}
549: %; solid line) and $R=\frac12 \sqrt{\pi/2}$ ($2g_d=g_{\rm 3D}$; dashed line).
550: %The depletion diverges at the point of instability of the condensate
551: %, which for $R=\sqrt{\pi/2}$ is at $A_c=3.446$. }
552: \end{figure}
553: \vspace*{-2.5em}
554: \end{center}
555:
556:
557:
558: %, with strong dipole-dipole interaction induced quantum correlations.
559:
560:
561: In conclusion, we have shown that
562: dipolar quasi-2D Bose-Einstein condensates are extremely
563: stable systems as compared to their 3D counterparts.
564: This offers the potential of approaching, starting
565: from the mean-field physics of condensates, a strongly correlated
566: regime of dilute atomic/molecular gases with long-range interactions.
567: A conceivable experimental procedure is to
568: start from a quasi-2D dipolar condensate, and, by
569: decreasing $\omega_z$, to enter the crossover regime to 3D,
570: where the quantum depletion of the condensate becomes large.
571: %without endangering the stability by imaginary excitation energies.
572: If the dipole-dipole interaction dominates, i.e., in the limit $R\rightarrow
573: \sqrt{\pi/2}$, the quantity
574: $A/\sqrt 2= \rho (0)d_z^2 g_d = 2.4 \,{\rho(0)}/{\rho_c(0)}$,
575: where $\rho_c(0)$ is the critical 3D density for
576: Bogoliubov excitations above the mean-field condensate to be stable,
577: discussed in the above. Hence $\rho (0)d_z^2 g_d $
578: measures the diluteness of the system, that is the proximity to the
579: critical region.
580: Numerical values are
581: \begin{eqnarray}
582: \rho (0)d_z^2 g_d & = &
583: 5.5 \times 10^{-3} \alpha N_m^2\, \frac{\rho(0) [10^{14}
584: {\rm cm}^{-3}]}{\omega_z [2\pi\times {\rm kHz}]}\,,
585: \nonumber\\
586: \rho (0)d_z^2 g_d & = &
587: %4.3 \,\frac{\rho(0)}{\rho_c(0)}
588: 0.63 \alpha N_e^2 \,\frac{\rho(0) [10^{12}
589: {\rm cm}^{-3}]}{\omega_z [2\pi\times {\rm kHz}]}
590: \,,\label{diluteness}
591: \end{eqnarray}
592: in the magnetic and electric cases, respectively.
593: According to the first equation, %line of (\ref{diluteness}),
594: magnetically dipolar gases will allow for the realization of a purely
595: dipolar condensate, such that the depletion still remains small.
596: %, using, for example, the change of $g_{\rm 3D}$ incurred by
597: %a Feshbach resonance (for chromium demonstrated in \cite{Werner}) .
598: On the other hand, it is evident from the second line of
599: Eq.\,(\ref{diluteness}), that for electric dipoles
600: %a purely dipolar BEC, at least on the level of mean-field
601: %physics, will be difficult to realize, because
602: the densities at which strong depletion of the condensate sets
603: in are rather small, even for strongly increased axial trapping.
604: %Electric dipole-dipole interaction dominated gases
605: %will therefore enter a strongly correlated regime already at
606: %very moderate densities.
607: %, even though they can still be stable with respect
608: %to excitations above the Bogoliubov mean-field at these densities.
609:
610:
611: Among further lines of research offered by the stability potential
612: of quasi-2D dipolar condensates are the phenomena expected when they are
613: set in rotation.
614: %For instance, the vortex lattices in the condensed phase
615: %are altered by the dipolar interaction.
616: %and the quantum Hall states in the strongly correlated regime
617: When both contact and dipole interaction are present, various of
618: %single vortices, vortex lattice structure and fractional
619: %quantum Hall states of bosons
620: these phenomena have been explored
621: in \cite{Cooper}; quantum Hall states of purely
622: dipolar Fermi gases were studied in \cite{BaranovQH}.
623: The interplay of quasi-two-dimensionality
624: and rotation may generally
625: yield interesting new physics
626: %, as has been demonstrated for the contact interaction case
627: \cite{Sinha}.
628: It will, furthermore, be of interest to determine the change in the system's
629: stability region when the dipoles are no longer locked in one direction,
630: and thus to investigate the influence of spin waves on the stability of
631: %purely
632: dipolar %bosonic %quasi-2D
633: quantum gases of bosons.
634:
635:
636: I thank C. Zimmermann, N. Schopohl, L. Santos,
637: C. Iniotakis, and N.\,R. Cooper for helpful discussions.
638:
639: \begin{thebibliography}{499}
640: \bibitem{Griesmaier}A. Griesmaier,
641: J. Werner, S. Hensler, J. Stuhler, and T. Pfau,
642: %: {\em Bose-Einstein condensation of chromium}, %cond-mat/0503044
643: Phys. Rev. Lett. {\bf 94}, 160401 (2005).
644: \bibitem{Goral} K.~G\'oral, K.~Rz\c a$\dot {\rm z}$ewski, and T.~Pfau,
645: Phys.\ Rev.\ A {\bf 61}, 051601(R) (2000).
646: \bibitem{YiYouII} S. Yi and L. You,
647: Phys. Rev. A {\bf 63}, 053607 (2001).
648: \bibitem{SantosZoller} L. Santos, G.\,V. Shlyapnikov, P. Zoller,
649: and M. Lewenstein, Phys. Rev. Lett. {\bf 85}, 1791 (2000);
650: Phys. Rev. Lett. {\bf 88}, 139904(E) (2002).
651: \bibitem{ODell} D.\,H.\,J. O'Dell, S. Giovanazzi, and C. Eberlein,
652: %: {\em Exact hydrodynamics of a trapped dipolar Bose-Einstein condensate},
653: Phys. Rev. Lett. {\bf 92}, 250401 (2004).
654: %; Phys. Rev A {\bf 71}, 033618 (2005).
655: \bibitem{Baranov} For a review, see
656: M. Baranov, \mbox{\L}. Dobrek, K. G\'oral, L. Santos, and
657: M. Lewenstein, Phys. Scr. {\bf T102}, 74 (2002).
658: \bibitem{YiYou} S. Yi and L. You, Phys. Rev. A {\bf 61}, 041604(R) (2000).
659: \bibitem{Kotochigova}
660: S. Kotochigova, P.\,S. Julienne, and E. Tiesinga,
661: Phys. Rev. A {\bf 68}, 022501 (2003).
662: \bibitem{RotonDipolar} L. Santos, G.\,V. Shlyapnikov, and M. Lewenstein,
663: %: {\em Roton-Maxon Spectrum and Stability of Trapped Dipolar Bose-Einstein
664: %Condensates}, %cond-mat/0301474
665: Phys. Rev. Lett. {\bf 90}, 250403 (2003).
666: \bibitem{RotonsDipole} D.\,H.\,J. O'Dell, S. Giovanazzi, and G. Kurizki,
667: %: {\em Rotons in Gaseous Bose-Einstein Condensates Irradiated by a Laser},
668: %cond-mat/0308096,
669: Phys. Rev. Lett. {\bf 90}, 110402 (2003).
670: \bibitem{EPJD} S. Giovanazzi and D.\,H.\,J. O'Dell, Eur. Phys. J. D {\bf 31},
671: 439 (2004).
672: \bibitem{Holzmann} D.\,S. Petrov, M. Holzmann, and G.\,V. Shlyapnikov,
673: %: {\em Bose-Einstein Condensation in Quasi-2D Trapped Gases},
674: Phys. Rev. Lett. {\bf 84}, 2551 (2000).
675: \bibitem{Abramowitz} M. Abramowitz and I. Stegun (eds.), {\em
676: Handbook of Mathematical Functions}, Dover, New York (1970).
677: \bibitem{note} To find approximate analytical expressions for the spectrum,
678: note that the $w$-function is to a good approximation bounded
679: from above and below by $(z+\sqrt{z^2+2})^{-1} < \frac{\sqrt{\pi}}2
680: w(z)\le (z+\sqrt{z^2+4/\pi})^{-1} $ ($z\ge 0$) \cite{Abramowitz}.
681: \bibitem{Bogoliubov} N.\,N.~Bogoliubov, J.\ Phys.\ (USSR) {\bf 11}, 23 (1947).
682: \bibitem{Foldy} L.\,L. Foldy, Phys. Rev. {\bf 124}, 649 (1961);
683: investigated the consequences of Bogoliubov theory for a long-range
684: (specifically a Coulomb) potential.
685: \bibitem{Hohenberg} P.\,C. Hohenberg,
686: %: {\em Existence of Long-Range Order in One and Two Dimensions},
687: Phys. Rev. {\bf 158}, 383 (1967).
688: \bibitem{Fischer} U.\,R. Fischer,
689: %{\em Existence of Long-Range Order for Trapped Interacting Bosons},
690: Phys. Rev. Lett. {\bf 89}, 280402 (2002).
691: \bibitem{PitaLongRange} L. Pitaevski\v{\i} and S. Stringari,
692: %: {\em Uncertainty principle and off-diagonal long-range order
693: %in the fractional quantum Hall effect},
694: Phys. Rev. B {\bf 47}, 10915 (1993).
695: \bibitem{Schutzhold} R. Sch\"utzhold, M. Uhlmann, Y. Xu, and U.\,R. Fischer,
696: %: {\em Mean-field expansion in Bose-Einstein condensates with finite-range
697: %interactions}, {\sf preprint}
698: cond-mat/0505074.
699: %\bibitem{Werner} J. Werner {\it et al.}, Phys. Rev. Lett. {\bf 94},
700: %183201 (2005).
701: \bibitem{Cooper} N.\,R. Cooper, E.\,H. Rezayi, and S.\,H. Simon,
702: %{\sf preprint}
703: Phys. Rev. Lett. {\bf 95}, 200402 (2005);
704: %; J. Zhang and H. Zhai, Phys. Rev. Lett. {\bf 95}, 200403 (2005);
705: %E.\,H. Rezayi, N. Read, and N.\,R. Cooper,
706: %: {\em Incompressible liquid state of rapidly-rotating bosons
707: %at filling factor 3/2},
708: %cond-mat/0507064; J. Zhang and H. Zhai, cond-mat/0506118;
709: S. Yi and H. Pu, cond-mat/0509246.
710: \bibitem{BaranovQH} M.\,A. Baranov, K. Osterloh, and M. Lewenstein,
711: %: {\em Fractional Quantum Hall States in Ultracold Rapidly Rotating Dipolar
712: %Fermi Gases},
713: Phys. Rev. Lett. {\bf 94}, 070404 (2005).
714: \bibitem{Sinha} S. Sinha and G.\,V. Shlyapnikov,
715: Phys. Rev. Lett. {\bf 94}, 150401 (2005).
716: \end{thebibliography}
717: \end{document}
718:
719:
720: The numerically calculated dependence of $A_c$ on $R$ is
721: shown in Fig.\,\ref{AofR}.
722: \vspace*{-0.5em}
723: \begin{center}
724: \begin{figure}[t]
725: \psfrag{2/3}{\normalsize $\frac23$}
726: \psfrag{Ac}{\normalsize $A_c(R)$}
727: \psfrag{R}{\normalsize $R/R_{\rm max}$}
728: \centerline{\epsfig{file=Fig3.eps,width=0.35\textwidth}}
729: \caption{\label{AofR}
730: The critical value of $A=A_c$, above which the in-plane Bogoliubov
731: spectrum (\ref{Bogoliubov}) acquires imaginary parts,
732: as a function of $R/R_{\rm max}$, where $R_{\rm max}=\sqrt{\pi/2}$.}
733: %note the logarithmic scale on
734: %the $A_c$ axis.}
735: \end{figure}
736: \vspace*{-2em}
737: \end{center}